BGP aggregation

Unanswered Question
May 6th, 2010

Good day colleagues,

our customer(as4444-sample) is asking us to do aggregation for him and it should looks like he has done it.

is it possible? how to do it?

he advertises 2 network :

A.B.23.0/24 as-path: 4444 4444

A.B.24.0/24 as-path 4444 4444

we need to aggregate it to A.B.23.0/23 with AS-PATH: AS4444 {44444, 4444}

than our peers can see it as: A.B.23.0/23 as-path:  5555 4444 {4444, 4444}  (where 5555 - our as).

please tell me how to do.  now i have the config:

router bgp 5555
aggregate-address A.B.23.0 255.255.253.0 as-set attribute-map TEST_AGGREGATION
!
!
route-map TEST_AGGREGATION permit 100
set metric 0
set local-preference 400
set community 5555:20
!

I was trying to put a prepend(4444) into TEST_AGGREGATION, but it is not working...
% "TEST_AGGREGATION" used as BGP attribute route-map, set as-path prepend not supported

--

Thank you for any comments

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
patrickvanham Thu, 05/06/2010 - 01:52

If the third octet is 23 and 24 respectively you cannot aggregate that into a /23. The first aggregate to hold both ranges would be A.B.16.0/20 although this is quite a lage aggregate and may overlap other ranges. Two /24 networks is the best you can do given a thord octet of 23 and 24 resp.

kozorezdi Thu, 05/06/2010 - 03:20

sorry, i was too lazy to count it (it is a sample)...

it is not a problem could be like this:

A.B.22.0/24 as-path: 4444 4444

A.B.23.0/24 as-path 4444 4444

to A.B.22.0/23

milan.kulik Thu, 05/06/2010 - 13:09

Hi,

this might help you:

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094826.shtml#changingtheattributesoftheaggregateroute

IMHO, you could add one

set as-path prepend 4444

to another outgoing route-map when sending the prefix to the neighbor (to prepend just for the aggregated prefix).

That way the AS-PATH might look like 5555 4444 (4444)

What I don't fully understand though, is: Which BGP attribute keeps the info "aggregated by ..."?

I'm afraid this would always be your router ID, not the customer router's one.

BR,

Milan

kozorezdi Fri, 05/07/2010 - 02:18

Hi Milan,

it looks like i can't hide my aggregation.  when i added a prepend for this prefix into out-route-map for all my iBGP peers, on the remote side i don't see it prepend... (clear ip bgp soft out - done)

--

Have a nice day,

Dmitry

milan.kulik Mon, 05/10/2010 - 04:37

Hi Dmitry,

I'm confused by the iBGP involved (outbound prepending does not work for iBGP neighbours, of course).

I'm not able to follow your target, I'm afraid. Could you pls explain once more with details?

Maybe prepending used within an inbound route-map on your router might help?

BR,

Milan

kozorezdi Mon, 05/10/2010 - 21:46

Good morning Milan,

On a router I have a session with a customer. I am doing an aggregation of the customer's prefixes on this router.

All other routers with eBGP connection have to send this aggregated prefix to peers, upstreams.

I am looking for any solution to do some modification only on the router with client's connection and provide this info to all other iBGP routers. That is why I was trying to prepend AS on iBGP session ( I was thinking we have some exclusion for aggregated prefixes - funny idea To do modification on edge routes it is not an option.  That is why I don't see a solution for this task.  Thank you for your help.

--

Dmitry

Actions

This Discussion